First of all, I appreciate the desire to open such a group.
I have some feedback to this, some pretty fundamental. My core criticisms:
- The concept of "learnability" is ill-defined and generally one I've frequently criticised here. This would make the first effort of the group a definition of its topic.
- The artifacts of this group are rather unclear, it explicitly avoids producing content. I think a different scope would enable a better artifact.
- The group wants to provide output to educators, but the only group directly mentioned in the "Call To Participation" are data scientists.
- It focuses on inspecting the language rather than the teaching practice.
- I recommend to not lean on the concept of a "Learning Style", which has not held in multiple studies in the last decades(!). "Learning Styles" are often a proxy for certain things missing, e.g. we have almost no visual tools for explaining memory - which might be flaw for everyone that some attribute to their "style", while it's actually useful to many.
While I appreciate the desire for a user study, the subject matter, especially around programming languages is what I call a research problem, which does not seem to be a good first task for a group. It is very hard to assess and de-bias such a dataset. Self-Reporting is also highly problematic for detailed conclusions.
I think making the language "better learnable" is a good desire, but there's lower hanging fruit. My favorite example here is the
turbofish-Operator: while it is a wart in the language and might trip people, it has no interdependence with other parts of the language and can just be learned as an odd fact. Science Educators teach such facts easily every day. The other side is that Rust has a number of fundamental concepts that are very new to people - for example: is Ownership or Borrowing the dominant concept in Rust? How do we get people to model with Ownership quick?
Honestly speaking, on almost every RFC where "learnability" ended up the final argument, my assessment as an educator was that the change was not worth the effort and cost of backwards compatibility.
We do, however, have a pool of knowledge already within the Rust project even: by speaking directly to educators and collecting practices. We do have people that taught Rust over the last 5 years, in settings from small to large groups. The have already tested many approaches. I think it's a way faster path to success for such a group to start speaking directly to or build it out of those people and have them condense their experiences into supplement material. Rather than changing "the book", producing co-material for those interested in teaching would be a huge boost to the ecosystem. Also, it can provide a number of sources to those with the desire to create courses.
This has multiple advantages:
a) It draws educators into our project and gives them a place. This forum can also be used for feedback.
b) It makes the group operate on what is currently available rather then having the group set up a foundation first.
c) It is common practice in many organisations that revolve around training/teaching, e.g. every sports association as a trainers forum rather than a "what do we change about the sport to make it easier?".
Currently, the effort of producing teaching Rust means reinventing the world every time - addressing this may yield much better results much faster.
: In that academia is still running studies for assessing those topics.